Introduction

Marburg virus disease (MVD) is a severe disease caused by the Marburg virus (MARV), with a case fatality rate varying between 23 and 90%, depending on the outbreak1,2. MARV is transmitted to humans from Egyptian fruit bats (Rousettus aegyptiacus)2. Subsequently, MARV is spread between humans through infected blood, body fluids, or contaminated surfaces3. Currently, there are no licensed treatments or vaccines for MVD. Only management of the symptoms is possible, with a focus on hydration and replenishing depleted electrolytes4. MVD patients initially present with flu-like symptoms with various clinical presentations including chills, fever, headache, sore throat, myalgia, and joint pains. As the disease progresses, the MARV infection may worsen, leading to internal hemorrhage, multiple organ failure, and death3,5. It has been reported that MARV impairs the antiviral immune responses by dysregulating mononuclear phagocytic cells, including monocytes, macrophages, Kupffer cells, and dendritic cells6. Subsequently, virions disseminate in the body and induce aberrant upregulation of cytokines and chemokines, leading to vascular permeability and coagulation abnormalities7. Lastly, the viruses may impair the functions of various organs such as the liver and the kidneys8. Despite this background, the pathogenesis of MVD is still not fully understood, and there is a paucity of data on biomarkers that can be used as monitoring tool(s) for MVD patients. Previous studies on MVD-induced biomarkers reported an increase in liver enzymes, alanine aminotransferase (ALT), aspartate aminotransferase (AST), and creatinine levels in MVD patients5. Other findings postulated lymphopenia, thrombocytopenia, and disseminated intravascular coagulation in the first weeks of MARV infection8,9. Previous findings on MVD biomarkers were derived from a limited number of patients during the very first MVD outbreak9,10. To comprehensively understand MVD-specific biomarkers, animal models were suggested, but only non-human primates (NHP) were susceptible to Marburg viruses. MARV-infected NHP displayed hallmark pathological features of MVD disease, similar to those observed in humans. Findings showed increased liver transaminases and bilirubin, indicating damage to the liver and kidneys11,12,13,14. In the same models, upregulation of pro-and anti-inflammatory cytokines was noted15. Data from MARV-susceptible mice infected with MARV-Angola and Ravn virus revealed elevated ALT and Alkaline phosphatase (ALP), amylase and total bilirubin levels16. In addition, mouse-adapted Ravn strain was also suggested for deciphering pathogenesis associated with MVD17. Other studies explored the biomarkers associated with MARV infection in Guinea pigs18 and MARV Syrian Golden hamsters, but these models did not adequately reflect human MVD19. Although some biomarkers were observed in certain animal models but it is likely that the mutated-genome viruses may not replicate the human infection. Notably, among the tested models, only NHP could reproduce the dynamics of biomarkers observed in MARV- infected patients.

The pathogenesis of MARV remains elusive, and data from MVD patients are limited due to its high fatality rate. The survival rate of MVD patients in a recent outbreak in Rwanda was 77% (n = 66)20. This is the largest number of MVD survivors ever documented since the first occurrence of MVD outbreak. These data provide an opportunity to comprehensively understand the kinetics of biosignatures associated with MVD. The present study hypothesized that the selected biomarkers may be used to monitor MVD progression. To address this question, we performed a retrospective analysis of hematological and biochemical markers collected from MARV-infected patients.

Results

Selected hematological and biochemical factors are modulated in MVD patients and normalised in recovered individuals

Critical biomarkers for monitoring MVD patients were regularly measured from the first days of admission until recovery. The analyses included data ranging from day 1 to day 21. Our findings indicated that biosignatures for liver and kidney function, such as AST, ALT, and creatinine, were increased in the first days of infection, and gradually normalized in the recovery phases (Fig. 1a–c). In addition, data revealed fluctuations in electrolytes (K+, Na+, and Cl−), throughout the course of MARV infection (Fig. 1d–f). Further analyses were performed to compare biological markers between the day of admission and the day of discharge (recovered patients). Key markers, such as ALT, AST, and creatinine were significantly lower in MVD recovered subjects as compared to newly admitted patients (p < 0.0001 each) (Fig. 1g–i). In addition, electrolytes, including Cl−, K+ and Na+, were notably increased in MVD-recovered patients (p < 0.005, 0.0001 and 0.0004, respectively) (Fig. 1j–l).

Fig. 1
figure 1

Kinetics of biochemical parameters in MARV-infected patients and recovered individuals. (a–f) Multiple line graph shows the levels of Alanine aminotransferase (ALT), Aspartate amino transferase (AST), creatinine, K+, Na+ and Cl− electrolytes measured from day 1 (d1) until the discharge of the MVD patients. Data were analysed up to 21 days. (g–l) The boxplot indicates the comparison of biomarkers between the day of admission (d1) and the day of discharge (recovered patients). Wilcoxon signed-rank test was used for comparisons and p-values were calculated. Significant differences are indicated as: *** < 0.001, ** < 0.01, and * < 0.05.

Full size image

MARV infection also induced dysregulation of hematological parameters, and most were restored in recovered individuals (Fig. 2a–f). Platelets, lymphocytes, and white blood cells, were significantly increased in MVD-recovered subjects (Fig. 2g–i). No significant changes were noted in basophils and monocytes (Fig. 2j–k). Eosinophils were elevated in MVD recovered individuals (Fig. 2l), and our findings also indicated a downward trend in neutrophils (Suppl. Fig. 1a). Hemoglobin, hematocrit levels, and the number of red blood cells were slightly decreased in MVD-recovered individuals (Suppl. Fig. 1b–d). However, hemoglobin concentration and the size of red blood cells hardly changed, as indicated by mean corpuscular volume and concentration (Suppl. Fig. 1e–g). Our data did not suggest any change in blood urea concentration (Suppl. Fig. 1h).

Fig. 2
figure 2

Kinetics of hematological parameters in MARV-infected patients and recovered individuals. (a–f) Multiple line graph shows the number of platelets and white blood cells, frequencies of monocytes, lymphocytes, basophils and eosinophils measured from day 1 (d1) until the discharge of the MVD patients. (g–l) The boxplot indicate the comparison of various immune cells between the day of admission (d1) and the day of discharge (recovered patients).Wilcoxon signed-rank test was used for comparisons and p-values were calculated. Significant differences are indicated as: *** < 0.001, ** < 0.01, and * < 0.05.

Full size image

The kinetics of biochemical and hematological markers discriminate between MARV-infected and recovered individuals

To investigate the key biomarkers for monitoring MVD, we initially classified the study subjects into three groups according to the day of discharge: early (< 5 days), middle (5–10 days), and late recovery (> 10 days). The analyses focused on the levels of biomarkers on the day of admission (denoted as MVD patients) and the day of discharge (denoted as recovered subjects). Principal component analysis (PCA) was performed to identify the clusters of correlated variables. The findings showed a difference in biochemical and haematological markers between MARV-infected patients and those in recovery phases; however, the latter were hardly distinguishable (Fig. 3a). Liver enzymes (ALT, AST), creatinine, red blood cells(RBC) and hematocrit (HCT) were highly elevated in MARV-infected patients, while platelets(PLT), electrolytes (Na+, K+), and lymphocytes were enriched in recovered patients (Fig. 3a,b). These markers were considered as potential candidates for discriminating between the study clusters. Next, we compared the concentration of ALT, AST, and creatinine in early, middle, and late recovered individuals and MVD patients. ALT levels could distinguish all study clusters; MVD patients and middle recovery (p = 0.0001), late and middle recovery (p = 0.0002), early and middle recovery (p = 0,0007) (Fig. 3c). Similarly, AST levels differentiated the groups; MVD patients and middle recovery (p = 0.0001), late and middle recovery (p = 0.001), early and middle recovery (p = 0.003) (Fig. 3d). On the other hand, creatinine concentration only discriminated between MVD patients and recovered subjects (Fig. 3e). Analyses also showed that MVD patients and recovered patients could be differentiated by lymphocyte and platelet levels (Fig. 3f,g). RBC and HCT markers did not differentiate the study clusters (Suppl. Fig. 2a,b).

Fig. 3
figure 3

Dynamics of biomarker candidates for monitoring MVD progression. (a) The principal component analysis (PCA) depicts the clusters of early recovered patients (< 5 days), middle (5–10 days) and late (> 10 days) recovered patients. (b) A PCA analysis correlated and clustered all investigated biomarkers. (c–g) Biomarker candidates AST, ALT, creatinine, platelets and lymphocytes were compared between the defined phases of recovery (day of recovery) and MVD patients (day of admission). Kruskal–Wallis test was used to compute stastical differences. p-values were calculated. Significant differences are indicated as: **** < 0.0001,*** < 0.001, ** < 0.01, and * < 0.05.

Full size image

Liver enzymes (ALT,AST), creatinine, lymphocytes and platelets are key biomarkers for monitoring MVD progression in MARV-infected patients

To confirm the capacity of AST, ALT, creatinine, lymphocytes, and platelet markers to discern MVD phases, receiver operating characteristic (ROC) regression was performed using the early recovery phase as a control. The ALT marker was effective in discriminating between middle, late recovery, and MVD patients(AUC: 0.86, 0.90, 0.94, respectively) (Fig. 4a). Likewise, AST also showed a strong discriminatory power for separating middle, late recovery, and MVD patients (AUC: 0.81, 0.95, 0.94, respectively) (Fig. 4b). Similarly, creatinine could distinguish the MVD patients and MVD recovered subjects (AUC: 0.94), but findings on late recovery were not conclusive (Fig. 4c). Moreover, platelets accurately discriminated between middle, late recovered individuals, and MVD patients (AUC: 0.61, 0.79, 0.90, respectively) (Fig. 4d). Finally, lymphocytes could only differentiate between middle and late recovered subjects (AUC: 0.56, 0.77, respectively) (Fig. 4e). Finally, the levels of the identified biomarkers were compared (day of admission only), and besides creatinine, other markers did not show any significant difference (Suppl. Fig. 3a–e). Paired analyses of the days of admission and recovery revealed that most biomarker candidates reverted to normal in early and middle recovered patients (Suppl. Fig. 4a–e, left and middle) but ALT, AST and lymphocytes were not normalized in late recovered individuals (Suppl. Fig. 4a–e, right). Lastly, we analysed the identified biosignatures in reference to physiological ranges, most markers were dysregulated in newly admitted patients and normalized in MVD-recovered individuals (Suppl. Fig. 5a,b).

Fig. 4
figure 4

Sensitivity of the identified markers for discriminating between MARV-infected individuals and recovered patients. (a–e) Receiver operating characteristics (ROC) regression analysis was performed for all identified biomarkers for monitoring MVD such as ALT, AST, creatinine, platelets and lymphocytes respectively. Sensitivity and Specificity of the biomarkers were computed at 95% C.I, AUC and cutt-offs were calculated. AUC \(\ge\) 0.7 was considered as a powerful discriminator.

Full size image

Discussion

The pathogenesis of MVD remains poorly understood, and since the fatality rate of MVD is up to 90%, studies investigating the kinetics of biosignatures associated with MARV infection are limited. To our knowledge, this is the first study of this nature with a large sample size. Our findings provide novel insights into hematological and biochemical biomarkers that can be useful for monitoring MARV infection. Initially, we investigated the kinetics of various biological signatures during MARV infection. Several markers were dysregulated in the earlier days of the disease and adjusted in later stages. Notably, this study showed that liver enzymes (ALT, AST) and creatinine were elevated in the first days of MVD, and normalised during the recovery. Similar findings were reported during the outbreak of MVD in Germany9, and the Ebola epidemics21,22. Consistent findings of increased liver transaminases and creatinine in filovirus-infected individuals support the previous reports suggesting MARV or Ebola viruses to attack hepatocytes and kidney cells23,24. Since the liver is a source of proteins, including coagulation factors and antiviral molecules (i.e., interferons), MARV replicates in liver cells and affects their functions. This viral replication leads to significant damage and death of liver cells. The extensive damage to liver tissue disrupts vital liver functions, including the synthesis of clotting factors and metabolic functions. Currently, viral load can be used as a tool for predicting the severity of the viral diseases, and depletion of viruses may correlate with the metabolic status of the MVD patients25. We reasoned that some biological markers can be instrumental for monitoring MVD. Our findings pointed out ALT, AST, creatinine, platelets, and lymphocytes as potential biomarker candidates for discriminating between early, middle, and late recovered individuals. Since varying levels of biomarkers on the day of discharge may have been influenced by the initial levels of the markers (day of admission), we further interrogated our data to exclude possible confounding factors. Comparison of the initial values among early, middle, and late recovered patients did not suggest any significant difference in AST, ALT, platelets, and lymphocytes, suggesting dynamics of biomarkers throughout the disease. Further analyses indicated that the ALT, AST, creatinine, platelets, and lymphocyte biomarkers were restored slowly over the course of the disease in late recovered subjects, postulating a correlation between MARV suppression and identified biomarkers. These findings are intriguing, since further ROC analyses hardly discriminated between late recovered subjects and MVD patients, reinforcing ALT, AST, creatinine, platelets, and lymphocytes as key factors for monitoring MVD. In line with our findings, liver transaminases were recently proposed as potential prognostic determinants for SARS-CoV-2 and influenza viruses26,27,28. Our analyses did not replicate the findings that postulated urea and bilirubin as factors that predict viral disease progression. The observed discrepancies may be due to the nature of the samples used, and it is also well known that urea nitrogen can be influenced by external factors29. There are multiple lines of evidence suggesting dysregulation of immune responses in MARV patients. Lymphocytes are vital cells in viral infections, but published data could not elucidate their functions in MVD23. Notably, our analyses indicated increased lymphocytes in early and middle recovered patients, positing that the lymphocytes may induce anti-MARV immunity. However, a previous study reviewed the role of cellular immunity in contributing MVD, and it was noted that MARV may induce exhaustion and apoptosis in T-lymphocytes23. Since T-cells play vital roles in suppressing viruses, MARV may induce evasion mechanisms in humans such as the upregulation of inhibitory molecules that prevent effective recognition of MARV by immune cells.

Our data also revealed a significant decrease in platelet levels in MARV patients that were restored in recovered individuals. Although only 3% of MVD patients exhibited bleeding as previously reported30, a depletion of platelets may lead to increased hemorrhage in MVD patients. Likely, MARV may directly kill the platelets, or antibodies against platelets or their precursors can be induced31. This study poses some limitations; MVD patients can present various clinical stages on the day of admission and this may affect the classification of study clusters. Therefore, our data may not represent the entire spectrum of MVD from the day of infection to recovery. All findings were interpreted in reference to the day of admission. In the future, well-designed mathematical models can be utilized to analyse multiple parameters from the day of MARV infection to recovery. Reliable animal models can also shed light on MVD-induced biosignatures.

In conclusion, our study identified the key biomarkers for monitoring MARV infection. The markers for liver and kidney functions, ALT, AST, and creatinine, are paramount for tracking MVD development. Lymphocytes and platelets are also crucial indicators for MVD progression. Prospective studies can explore the usefulness of the identified markers in clinical settings, especially for the management of deadly viruses.

Materials and methods

Study participant characteristics

In total, we analysed retrospective data from 51 MVD patients [Age: 21–51], including 15 females and 36 males. All hematological and biochemical data were collected during the MVD outbreak in Rwanda (September 2024–December 2024). This study was approved by the Rwanda National Ethics Committee, under the clearance number: 121/RNEC/2024, and all methods were performed per relevant guidelines and regulations. Due to the retrospective nature of the study, the Rwanda National Ethics Committee waived the need to obtain informed consent. During the MVD outbreak, the MARV was detected using the Altona RealStar® Filovirus Screen RT-PCR Kit 1.0. A cycle threshold (CT) value of ≤ 36 was considered positive. To comprehensively investigate the prognostic biomarkers and also take into consideration MVD development, we defined the recovery phases: early (< 5 days, middle (5–10 days), and late (> 10 days). In this study, MVD recovered individuals are defined as the patients who were discharged from the treatment center after two consecutive negative PCRs for MARV.

Biochemical and hematological analyses

This is a retrospective study that analysed hematological and biochemical biomarkers collected during the MVD outbreak. Both hematology (Sysmex Xs) and biochemistry (CobasC111, Roche diagnostics) analysers were used and consistently calibrated as stipulated by the manufacturers’ guidelines. Various biological factors were measured to regularly assess liver and kidney functions, blood coagulation abnormalities, as well as immune responses to MARV infection. The analyses included the following markers: Aspartate aminotransferase (AST), alanine aminotransferase (ALT), urea, creatinine, white blood cells (lymphocytes, basophils, eosinophils, neutrophils and monocytes), red blood cells, hemoglobin, mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), and platelets. All values were interpreted as indicated in the manufacturer’s protocol. AST: Men 16–47 U/L, Women 16-32U/l; ALT: Men 12-43U/L, Women 12- 33U/L; Creatinine: 0.5–1.1 mg/dl; Lymphocytes 20–40%, Platelets: 150.000–450.000/uL. Data were recorded and kept in the data management system of the Rwanda Biomedical Center, Ministry of Health.

Graphical depiction and statistics

Data were retrieved from the data management system of the Rwanda Biomedical Center and analysed anonymously. All analyses were of a summary nature, and no individuals could be identified. Wilcoxon signed rank test, Mann–Whitney, or Kruskal–Wallis tests were used to compare groups accordingly. A p-value < 0.05 was considered significant. To investigate biomarkers with high discriminatory power, firstly, a principal component analysis (PCA) was used to cluster all possible correlated variables. FactoMineR, factor extra, and ggforce packages were used. Further analyses suggested potential biomarkers, which were validated by receiver operating characteristic (ROC) regression analyses. Early recovered subjects served as a control in the ROC analyses. All statistical analyses were performed using the GraphPad software version 10.4.1(2024) and R software version 4.3.2 (2023-10-31).